Keywords: model-based reasoning, structural identification, data mining, reliability.

This chapter describes informatics research that aims to increase the usefulness of performance monitoring through new methodologies for data interpretation. Following development of new sensor technologies as well as increases in data storage capacity, data interpretation is the most resilient barrier to widespread use of sensors on and in civil infrastructure. A model-based approach to data interpretation is part of structural-identification methodology and in this chapter aspects of advanced reasoning methods are described. More specifically, topics include multi-model reasoning, stochastic search, data-mining and reliability analysis. Results are then placed in the context of research into model-free structural identification through description of aspects of a state-of-the-art report that has been prepared within the Structural Identification Committee of the American Society of Civil Engineers.

The number of opportunities for enhancing infrastructure management through advanced informatics is increasing. Model-based identification methods can be categorised into five orders according to the number of models that are considered, the inclusion of errors explicitly in the methodology, the use of probability analysis and finally, application of data mining methods. Some methods, such as those employed in Structural identification Orders 2, 3 and 4 would not be possible without advanced informatics. Each method is useful for particular situations and engineers should use high order methods only when they can be justified through the possibility of higher quality support. Such high-level support requires an accurate knowledge of aspects such as measurement and modelling errors and this is not always available.

Certain signal analysis techniques, when adapted for infrastructure management tasks, can be used to detect anomalies in structures without behaviour models. These are called model-free methods. Many methods that have been proposed in the literature are not successful when evaluating harmonics that have periods ranging from less than a second to one year - as is the case for civil engineering infrastructure such as bridges. Further challenges include accommodating missing data and detecting outliers. Model-free methods and model-based methods are complementary because they are used in different practical contexts.

The field of advanced informatics is expanding the range of support that is available to engineers who are responsible for infrastructure management. As the environments where complex infrastructures are present become increasingly uncertain, the need for such support is expected to increase.

The full-text of this chapter is not yet available.

go to the previous chapter
go to the next chapter
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)